Background
Stomach adenocarcinoma (STAD) is the fifth most common cancer type worldwide and the third leading cause of cancer-related deaths [
1]. STAD is the most common histological type of malignant tumor that originates in the stomach and is a heterogeneous disease with different phenotypes and genotypes. Although the treatment of STAD has rapidly advanced due to the development of laparoscopic technology [
2], because of the absence of clear early symptoms, most patients with STAD are already at an advanced stage at the time of diagnosis and are prone to distant metastasis; thus, the prognosis remains poor [
3‐
5]. Therefore, new STAD treatments and prognostic targets are urgently needed to improve the survival rate of these patients.
The ceRNA hypothesis [
6] was first proposed in 2011, and posits that lncRNAs regulate the expression of target mRNAs by adsorbing miRNAs and thereby act as ceRNAs; they competitively bind to shared miRNAs, inhibiting the degradation of mRNA and thus acting as miRNA sponges. To date, the complex oncogenesis-related ceRNA network of lncRNA–miRNA–mRNA interactions has been explored in various types of cancer, such as colorectal cancer [
7], cervical cancer [
8], and lung squamous cell carcinoma [
9]. However, ceRNA network analysis in patients with STAD is relatively rare.
Ferroptosis was first proposed as a new form of cell death in 2012. Ferroptosis leads to cancer cell death by regulating iron-, amino acid and glutathione-, and ROS-metabolism, especially for the removal of aggressive malignancies that show resistance to conventional therapies [
10]. Ferroptotic cancer cells may influence the therapeutic effect of anti-tumor immunity by releasing signals such as oxidized lipid mediators, or some iron-sagging cells may suppress the immune system and promote the growth of tumor cells [
11]. However, few studies have explored the correlation between ferroptosis and immune infiltration in STAD.
Therefore, we comprehensively analyzed and identified some RNAs, including lncRNAs, miRNAs, and mRNAs. Based on these RNAs, we constructed a ceRNA network to elucidate the lncRNA–miRNA–mRNA interactions in STAD and identified biomarkers for the development of therapies for STAD. Finally, ferroptosis-related genes were screened in the ceRNA network and subjected to differential expression and prognostic analyses, to explore the relationship between them and immune infiltration in STAD.
Methods
Data collection and preprocessing
We used the genomics data commons data transfer tool (
https://gdc.cancer.gov/access-data/gdc-data-transfer-tool.html) to download the published the cancer genome atlas (TCGA) RNA-seq data, miRNA data, and the corresponding clinical information on STADs. The screening criteria for lncRNAs and mRNAs included "Project: TCGA-STAD," "Experimental strategy: RNA-Seq," and "Workflow type: HTSeq-Counts" which included 375 STAD tissues and 32 normal gastric tissues. The screening criteria for miRNA included "Project: TCGA-STAD," "Experimental strategy: miRNA-Seq," and "Workflow type: miRNA Profiling", which included 446 STAD tissues and 45 normal gastric tissues. The clinical follow-up datasets from 409 patients with STAD were also obtained from TCGA database.
Analysis of the DE lncRNAs, miRNAs, and mRNAs
We used the “edgeR” package [
12] to screen DElncRNAs, DEmiRNAs and DEmRNAs with thresholds of false discovery rate (FDR) < 0.01 and |log 2 (fold change [FC])|> 1. Volcano plots were generated using the “ggplots.”
Prediction of DEmRNAs targeted by DEmiRNAs
Construction of the ceRNA network
Based on the DERNAs and the relationships between the identified miRNA-mRNA and miRNA-lncRNA pairs, Cytoscape (version 3.7.2) was used to construct and visualize the ceRNA network [
17].
Functional enrichment analysis
The “ClusterProfiler” software package [
18] in R software was used to perform Gene Ontology (GO) functional enrichment [
19] and Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analyses [
20]. P < 0.05, was used as the threshold of statistical significance in the GO and KEGG enrichment analyses. The results were visualized using the “ggplots” package of R software.
Screening for ferroptosis-related genes in the ceRNA network and prognostic analysis
Sixty ferroptosis-related genes were queried from the reported literature [
21‐
24] and are shown in Additional file
1. The ferroptosis-related genes were intersected with DE miRNA-targeted genes to derive the associated genes. For the selected genes, the median expression value was used as the cut-off point, and the patients with STAD were divided into high and low-expression groups, and Kaplan–Meier survival curves were drawn. The log-rank test was used to compare the difference in survival time between the high and low-expression groups. A similar analysis was performed for upstream miRNAs. Finally, we performed univariate and multivariate analyses of ferroptosis-related genes using the R package "survival" to identify their prognostic significance.
Gene set enrichment analysis (GSEA)
In the TCGA cohort, we divided the 375 patients with STAD into two groups according to the median expression values of ferroptosis-related genes and chose the h.all.v6.2.symbols.gmt in the Molecular Signatures Database (MSigDB) as the reference gene set to perform GSEA analysis.
CIBERSORT estimation and immune-related analysis
We used the CIBERSORT algorithm to evaluate 22 immune cell types in STAD. Samples were only used for further analysis when the CIBERSORT output p < 0.05. To show the correlation of various immune cells, a co-expressed heatmap was drawn based on the results of the Spearman correlation analysis. The Wilcoxon rank-sum test revealed statistically significant differences in the proportion of immune infiltrating cells between the two groups with high and low expression of ferroptosis-related genes (p < 0.05). Spearman correlation analysis was performed for the selected ferroptosis-related biomarker in the ceRNA network and the proportion of each related immune cell with p < 0.05. Immune cells differentially expressed in the high and low groups of ferroptosis-related genes were intersected with immune cells associated with the expression of ferroptosis-related genes using the R package “VennDiagram” to obtain immune cells associated with ferroptosis-related genes. Spearman correlation analysis was used to assess the correlation between ferroptosis -related genes and the expression of immune checkpoints PD-1, PD-L1 and CTLA4. Finally, we downloaded two immunotherapy cohorts, the IMvigor210 cohort of atezolizumab (anti-PD-L1 antibody) for advanced metastatic cell carcinoma [
25] and the GSE78220 cohort of pembrolizumab (anti-PD-1 antibody) for melanoma [
26]. The correlation of iron death-related gene expression with anti-PD-L1 and PD-1 treatment response was analyzed in these two immunotherapy cohorts, respectively, and P < 0.05 was considered statistically significant.
Cell lines and cell culture
STAD cell lines AGS, MGC-803, SGC-7901, BGC-823, MKN-45, MKN-28, HGC-27, and human gastric epithelial cells (GES-1) were purchased from ATCC (American Type Culture Collection, Manassas, VA, USA). All STAD cell lines were cultured in 1640 medium (Gibco, Gaithersburg, MD, USA) supplemented with 10% fetal bovine serum (FBS, Gibco-BRL, Paisley, UK), 100 U/mL penicillin, and 100 μg/mL streptomycin at 37 °C in 5% CO2.
RNA extraction and quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from cell lines using TRIzol® Reagent (Invitrogen, Carlsbad, CA, USA). Total RNA was reverse transcribed into cDNA using PrimeScript™ RT Master Mix (Takara, Dalian, China) and then used to perform qRT-PCR with SYBR® qPCR Master Mix (Vazyme, Nanjing, China). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the internal control for gene quantification. The 2
−ΔCT value was calculated for every sample and normalized to that of GAPDH. The primer sequences used for PCR are listed in Additional file
2.
Cell Counting Kit-8 assay
Cell Counting Kit-8 (CCK-8, Dojindo Laboratories Kumamoto, Japan) for cell proliferation analysis, according to the manufacturer's instructions. Cells were grown in each well of a 96-well plate at a density of 2 × 103 cells/well. Afterwards, 100 ul of CCK-8 solution (CCK-8 solution was prepared from 10 ul of CCK-8 reagent and 100 ul of culture medium) was added to each well at different time points (24 h, 48 h, 72 h and 96 h), and the absorbance was measured at 450 nm after incubation at 37 °C for 2 h.
Transwell assay
For cell migration assays, Transwell chambers (Corning, USA) with 0.8um pore size were placed in 24-well plates. In the lower chamber, 1640 containing 10% fetal bovine serum was added, and then MGC-803 cell line transfected with siRNA were inoculated with serum-free medium in the upper chamber at a density of 5 × 104 cells/well and incubated for 48 h at 37 °C. Cells that migrated to the lower chamber were fixed with 4% paraformaldehyde for 30 min and then stained with 1% crystal violet for 30 min. The unmigrated cells at the bottom of the chamber were gently wiped with a cotton swab and the stained cells that had migrated to the lower chamber were photographed with a light microscope.
Apoptosis detection by flow cytometry
MGC-803 cell line transfected with siRNA were collected, stained with FITC Annexin V and propidium iodide (BD, USA) and analyzed for apoptosis by flow cytometry (BD, USA). PI-negative and FITC Annexin V-positive cells identified apoptosis at an early stage, while late or already dead cells were positive for both FITC Annexin V and PI. The results were analyzed by FlowJo software.
Discussion
The development of gastroscopy has gradually increased the rate of diagnosis of STAD, but most patients with STAD are in the progressive stage when they are detected. Thus, surgery is not effective, and the available treatment options include chemo-, targeted-, and immune-therapies. The discovery and development of immunotherapeutic agents have brought significant survival benefits to patients with STAD and are increasingly challenging the traditional treatment paradigms involving chemotherapy and targeted agents [
27]. Therefore, there is a need to further understand immunotherapy-related genes as novel prognostic markers for STAD. In the present study, the ferroptosis-related gene SLC1A5 was identified as a potential prognostic biomarker for STAD, its upstream molecule miR-137 was explored, and the correlation between SLC1A5 and tumor-infiltrating immune cells and immune checkpoints in STAD was investigated. Finally, the predictive value of SLC1A5 in immunotherapy response was evaluated.
SLC1A5 is a cell surface transporter that mediates the uptake of neutral amino acids, such as glutamine [
28]. The intracellular glutamine pool plays a key role in the sustained activation of the mechanistic target of rapamycin complex 1 (mTORC1) signaling, in which mTORC1 is a major regulator of cell proliferation, apoptosis, and autophagy [
29]. In erastin and rsl3 induced iron death, glutamine input and metabolism induce lipid ROS production, which leads to cell death [
30]. miR-137 or the inhibitor GPNA inhibits SLC1A5 and thus strongly inhibits glutamine catabolism, leading to the death of ironophilic cells. SLC1A5-mediated glutamine transport plays a crucial role in tumor cell metabolism, proliferation, and ferroptosis; therefore, inhibiting SLC1A5 and thus blocking glutamine transport is one of the approaches to treat solid tumors. miR-137 has been reported to be significantly downregulated in melanoma [
31,
32], glioblastoma [
33], colorectal cancer [
34], and non-small cell lung cancer [
35] compared to adjacent normal tissues. SLC1A5 is a target of miR-137 and is expressed at elevated levels in melanoma [
36], neuroblastoma [
37], and prostate cancer [
38]. MiR-137 was negatively correlated with SLC1A5, suggesting that SLC1A5 is a key target of miR-137, inhibiting the growth of cancer cells. In the present study, SLC1A5 was found to be highly expressed in STAD tissues compared to normal gastric tissues, while miR-137 was expressed at low levels. In addition, the prognosis of patients with STAD was better with low miR-137 and high SLC1A5 expression, probably because low miR-137 expression attenuated the inhibitory effect on SLC1A5, thereby inhibiting the development and progression of STAD cells. This idea has not yet been suggested in any study.
Glutamine is essential for the immune system for terminally differentiated immune cells, such as neutrophils [
39], macrophages [
39‐
41], and activated lymphocytes [
42,
43]. During naive T cell activation, SLC1A5 is required for rapid glutamine uptake [
44], as it promotes cell growth and proliferation in T cell receptor (TCR)-stimulated mTORC1 activation [
45]. SLC1A5 deletion can have an impact on T-cell effector functions, with impaired differentiation of helper T cells to Th1 and Th17 subpopulations [
44]. Activated lymphocytes strongly utilize glutamine [
29,
42,
46‐
48]. mTORC1 plays an important role in metabolic reprogramming, which is essential for NK and T cell effector functions [
49‐
51]. And upregulation of the glutamine transporter SLC1A5 is key to mTORC1 activity [
44,
52‐
54]. c-Myc is essential for NK cell metabolism and T cell activation [
55,
56]. In T cells, c-Myc expression is required for the activation of induced glutamine hydrolysis, and glutamine uptake is critical for T cell proliferation [
57]. Glutamine uptake via SLC1A5 is required for c- Myc induction in cytokine-stimulated NK cells [
55]. Amino acid translocation upregulates c-Myc, while positive feedback stimulates SLC1A5 expression, maintains mTORC1 activity and supports c-Myc expression. Nevertheless, studies on the role of SLC1A5 in immune cells, which play a key role in suppressing tumor growth, are only beginning. T helper follicular cells from CD4+ T-cell subsets help B cells and induce antibody responses, thus playing an important role in anti-tumor immunity [
58]. In the present study, we found that the percentage of T helper follicle cells in the high SLC1A5 expression group was higher than that in the low-expression group, and the expression of SLC1A5 was positively correlated with the content of T helper follicle cells. Monocytes are major regulators of tumor development and progression [
59] and are also an important source of long-term tumor-associated macrophages (TAMs) and dendritic cells (DCs) that form the tumor microenvironment [
60]. Our results showed a higher percentage of monocytes in the SLC1A5 low-expression group and a negative correlation between the two. This explains why the SLC1A5 high expression group has a better prognosis: one of the reasons may be that the infiltration of these two immune cells plays a key role.
Glutamine addiction has been reported to be one of the targets for cancer treatment by inhibiting glutaminolysis or enzymes in the glutamine transporter [
61,
62]. The current research on SLC1A5 in gastric cancer treatment is also focused on glutamine metabolism. Targeting SLC1A5 in gastric cancer produces antitumor effects by inhibiting the mTOR/p-70S6K1 signaling pathway [
63], glutamine mediates gastric cancer growth, and the efficacy of targeted glutamine therapy is dependent on the different expression patterns of the glutamine transporter ASCT2 and glutamate synthetase (GS) in specific gastric cancer groups [
64], the new monoclonal antibody KM8094 has a very high therapeutic potential in targeting the neutral amino acid transporter ASCT2 [
65,
66]. However, there are no studies that have explored the relationship between the ferroptosis-related gene SLC1A5 and immunotherapy. Immunotherapy, a therapeutic approach that boosts the immune system with drugs to fight tumors, currently plays a key role in cancer treatment [
67]. Among them, immune checkpoint inhibitors (ICIs) targeting cytotoxic T-lymphocyte antigen-4 (CTLA-4) and programmed cell death protein-1 (PD-1) are promising and may play an important role in immunotherapy [
68]. PD-1 is a member of the CD28 family and is essentially a suppressor receptor expressed mainly on activated T cells, B cells, macrophages, regulatory T cells (treg), and NK cells [
69,
70]. It binds to two kinds of ligands, PD-L1 and PD-L2, which are mainly expressed in T cells, B cells, macrophages, and dendritic cells [
71‐
73]. Tumors cause excessive activation of the PD-1/L1 signaling pathway, which in turn reduces T-cell activation and antigen-specific T-cell immune response, and finally bypasses immune surveillance, thus promoting tumor growth [
69,
74,
75]. In the present study, we found that the expression of SLC1A5 in the TCGA-STAD cohort was positively correlated with the expression of PD-1 (PDCD1), but negatively correlated with the expression of PD-L1 (CD-274). Therefore, the better prognosis of patients with high SLC1A5 expression may be related to the reduced expression of PD-L1, resulting in fewer PD-1-binding ligands and thus a weaker immune escape effect. In contrast, immunotherapy targeting PD-1 may improve the prognosis of patients with STAD. We also confirmed the predictive value of SLC1A5 for immunotherapy response in an anti-PD-1 immunotherapy cohort. There was a significant difference in SLC1A5 expression between responders and non-responders to anti-PD-1 therapy. Although there is no published cohort of immunotherapy patients with STAD, the above results are still suggestive of SLC1A5 as a predictive marker in immunotherapy of STAD patients.
Our study is a prediction generated by preliminary data analysis and hypothesis testing, and therefore has some limitations. First, the TCGA-STAD cohort had limited number of patients and a larger sample size is required to obtain more reliable data. Second, the targeted inhibitory effect of miR-137 on SLC1A5 in STAD needs to be experimentally validated. Third, the role of SLC1A5 in regulating immune cell infiltration and immune checkpoints needs to be further investigated.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.